Hire a data architect

A data architect designs how data flows through your organisation: which sources it comes from, where it is stored, which layers process it and who has access to it. They then make sure teams build within that framework.

At Data Science Lab, an experienced colleague takes on this role, focusing on strategy, the platform or governance. You can see which roles we place on assignment on the overview page.

The architect decides on the structure: which platform to use, which layers to create and what conventions to set for naming and access. The engineer builds the pipelines and data models within that structure.

In a small team, one person often does both. As the platform grows or more teams start using it, it pays to make architecture a separate role, so someone maintains an overview while others build.

In an architecture decision record: a log of decisions. According to Microsoft, each record describes the context, the options considered, the outcome and the trade-offs, with a status such as proposed, accepted or superseded. If a decision changes, you write a new record that supersedes the old one.

That way, your team can later check why something was built as it was. The architect tests risky components in advance with a proof of concept.

That depends on what you need them for. For Fabric, Microsoft recommends a warehouse for SQL reporting with full transaction support and a lakehouse for large volumes of data, semi-structured data and machine learning. Both reside in OneLake in an open table format.

Databricks recommends the medallion architecture: bronze for raw data, silver for cleaned data and gold for models and reports. It is a recommendation, not a requirement. Read more about building a data warehouse.

You need to design access, lineage and ownership into the architecture; adding them later is difficult. In Databricks, Unity Catalog manages access rights, tracks lineage — where data comes from — and logs activity for audits. Objects have a three-level name: catalog, schema and table.

Who owns each definition is an organisational question. It belongs under data governance.

According to the Dutch government, the Dutch Tax Administration resumed enforcement against false self-employment on 1 January 2025. Whether someone is effectively an employee is assessed based on how they work in practice, including supervision, integration into the organisation and whether they operate as an entrepreneur.

An interim data architect hired through Data Science Lab is employed by us. If you hire a self-employed professional, that question is yours to address. Have HR, procurement or an adviser review your hiring policy.

The decision log, design documents and agreed conventions stay in your repository, not ours. Where appropriate, the architect works alongside an internal colleague who can take over the role afterwards.

If one of our colleagues is such a good fit that you would like to hire them permanently, we are open to discussing a transfer.

If you have one data source and a few dashboards, a dedicated architect is more than you need; an experienced engineer can make those decisions. If the architecture is already in place, you may need more people to build it out instead.

If you do not yet know what you want to achieve with data, start with a maturity assessment or an AI roadmap. Architecture without a goal becomes a diagram on the wall.